Models for maintenance optimization: a study for repairable systems and finite time periods

The problem of selecting a suitable maintenance policy for repairable systems and for a finite time period is presented. Since the late seventies, examples of models assessing corrective and preventive maintenance policies over an equipment life cycle exist in the literature. However, there are not too many contributions regarding real implementation of these models in the industry, considering realistic timeframes and for repairable systems. Modeling this problem requires normally the representation of different corrective and/or preventive actions that could take place at different moments, driving the equipment to different states with different hazard rates. An approach to pattern the system under finite periods of time has been the utilization of semi-Markovian probabilistic models, allowing later a maintenance policy optimization using dynamic programming. These models are very flexible to represent a given system, but they are also complex and therefore very difficult to handle when the number of the system possible states increases. This paper explores the trade-off between flexibility and complexity of these models, and presents a comparison in terms of model data requirements versus potential benefits obtained with the model